Wavelength division multiplexed (WDM) optical communication systems are known in which multiple optical signals or channels, each having a different wavelength, are combined onto an optical fiber. Such systems typically include a laser associated with each wavelength, a modulator configured to modulate the optical signal output from the laser, and an optical combiner to combine each of the modulated optical signals. Such components are typically provided at a transmit end of the WDM optical communication system to transmit the optical signals onto the optical fiber. At a receive end of the WDM optical communication system, the optical signals are often separated and converted to corresponding electrical signals that are then processed further.
In earlier systems, the modulators would often intensity or amplitude modulate one polarization of an optical signal in accordance with a known on-off keying format. Using “direct detection” at the receive end, transmitted data or information is sensed as changes in the light intensity of the optical signal. In later systems, optical signals having the same wavelength but different polarizations (e.g., transverse electrical or TE and transverse magnetic or TM) have been separately modulated and combined multiplexed on to an optical fiber. Such polarization multiplexed optical signals at different wavelengths can be supplied to the optical fiber in order to further increase capacity.
In addition to direct detection systems, optical systems have been deployed that implement coherent detection, in which the optical phase of a transmitted optical signal is modulated in order to carry data. Coherent detection systems are known to have a greater noise tolerance than direct detection-based systems.
A receiver in a coherent system typically includes a light source or laser, also referred to as a local oscillator. Incoming light of the received optical signal, which, if polarization multiplexed, may be split by a polarization beam splitter (PBS) into two orthogonal signals, having the TE and TM polarizations, respectively. Each signal output from the PBS may be combined with the light output from the local oscillator and may be passed through a 90-deg optical hybrid circuit. The optical hybrid circuit, in turn, outputs further optical signals to four pairs of photodiodes or balanced photodetectors, which, in turn, generate corresponding electrical signals.
The electrical signals, which are typically in analog form, are next supplied to an analog-to-digital converter (ADC) circuit, which operates at a sampling rate to generate a series of digital samples at periodic time intervals. Each sample includes a plurality of bits. The samples may then be supplied to a digital signal processor (DSP), which processes the samples to extract the data carried by the optical signals.
The optical signals may be subject to various impairments including chromatic dispersion (CD), polarization mode dispersion (PMD), and cross-phase modulation (XPM). Such impairments, however, can be compensated or corrected with a known equalizer that may be implemented with the DSP. The equalizer may include a finite impulse response (FIR) that multiples the samples or portions thereof by different coefficients or weights (also known as tap weights) and then sums the resulting products. The FIR filter may have two outputs, each corresponding to the first and second polarizations of the transmitted optical signal.
One example of an optical signal modulation format that has been implemented in coherent systems is DP-QPSK (dual-polarized quaternary phase shift keying) transmitter. Here, light in each of two orthogonal polarizations of the transmitted optical signal carries two bits of information per symbol interval by phase modulation to four phase states separated by π/2 radians. In another exemplary modulation format, known as DP-BPSK (dual polarized binary phase shift keying) light in each of the two orthogonal polarizations carries one bit of information per symbol interval by phase modulating to two phase states separated by TC radians. While DP-BPSK carries only half the information as DP-QPSK, DP-BPSK modulation is preferred on optical fiber links whose transmission characteristics or impairments (e.g., CD, PMD, and XPM) are insufficient to support DP-QPSK.
As noted above, coherent receivers typically include an equalizer to process the received signals. The equalizer may include a finite impulse response (FIR) filter that multiples the samples or portions thereof by different coefficients or weights (also known as tap weights) and then sums the resulting products. To aid in setting tap weights of the FIR filter, known or predetermined symbols or data sequences may be carried by the optical signal output from the transmitter, and the filter coefficients may be set so that data output from the receiver matches the known data sequences.
If system parameters change, however, new tap weights need to be calculated, therefore requiring transmission of additional predetermined sequences, which consumes capacity that may otherwise be used to transmit user data.
Alternatively, alternatively, “blind equalization” techniques have been developed that can set the tap weights without transmission of predetermined data sequences. In systems that transmit DP-QPSK modulated optical signals, one blind equalization method is based on a so-called “Constant Modulus Algorithm” (CMA).
For DP-QPSK, a weakness of CMA is that some solutions to its tap weight calculation result in the data that was sent on one of the two polarizations at the transmitter appearing on both of the outputs of the FIR filter. This is termed a “degenerate state”. One approach to avoid the degenerate state is described in a paper by Papadias et al. and is often referred to as Multi-User CMA (MU-CMA). However for DP-BPSK, MU-CMA does not ensure that the two outputs of the FIR filter are the un-corrupted data that was transmitted. Instead, each FIR filter output can be a linear combination of the signals transmitted on each polarization (and resembling QPSK). Accordingly, although MU-CMA is suitable for processing data carried by DP-QPSK modulated optical signal, MU-CMA may not be reliably used to process data carried by DP-BPSK modulated optical signals. Thus, the same processor running or programmed to carry out MU-CMA may not be used to process data carried by both DP-BPSK and DP-QPSK modulated optical signals.
Another solution would be to modify the characteristics of the transmitted signal to resemble DP-QPSK such that MU-CMA produces a satisfactory outcome, i.e., data carried by each polarization is correctly output, and to construct the receiver such that the tolerance to impairments is similar of the original DP-BPSK.
A potential candidate would be so-called “π/2-shifted BPSK”. Here, the carrier phase is increased or decreased by π/2 radians from one symbol to the next according to the sign of an incoming data bit of each symbol. On any symbol carried by each polarization, the two possible transmitted phases or phase states are separated by pi radians; same as BPSK—as needed for impairment tolerance. Over time, the phase will be at one of four phase states; same as QPSK. Both polarizations alternate between the different phases relatively frequently in accordance with a repeating short bit sequence or coding pattern. It has been observed, however, that tap weight control remains unsatisfactory with MU-CMA even when BPSK modulated optical signals are transmitted in this manner.